Financial risk management is important because it is a means to achieving return enhancement. The prevalence of benchmark-focused investment strategies that concentrate on tracking-error management makes it is all too easy to forget that risk management is not an end in itself.
Technology is the application of science. The science that risk management technology applies is called modern portfolio theory (MPT), which is important because it solves the problem of how to achieve a risk target by varying portfolio allocation. This is the method commonly known as mean-variance optimisation.
The goal of financial risk management is not to completely eliminate financial risk but rather to control the trade-off between risk and return. MPT allows investors to identify those risks for which they are not being compensated and selectively eliminate them.
The investment cycle begins with investment research and market analysis. At this stage the focus is on return forecasting and risk estimation. The sensitivity of mean-variance optimisation to the relative magnitudes of return forecasts is widely recognised as a problem, which is commonly dealt with by using shrinkage techniques that scale down more extreme forecasts.
Risk estimation is more complex. The use of historical estimates from as long a period as possible has the merit of capturing many market events and scenarios, but has the drawback that it smoothes out the effect of turbulence. However, recently developed technology is available that addresses the need to stress test portfolios under turbulent market conditions.
It is a common complaint of users of mean-variance optimisation that it produces extreme allocations that can deviate drastically from their current holdings and from their benchmarks. When this happens, the customary response is to introduce constraints into the optimisation. Constraints are arbitrary and for a given level of risk a constrained solution offers a lower expected return than its unconstrained counterpart. Research into the use of risk tolerances has shown that by expressing a tracking error tolerance, investors are able to obtain less extreme allocations without resorting to arbitrary allocation constraints. It is also a lot less work for an investment manager to identify one or two risk tolerance numbers than to come up with dozens of pairs of minimum and maximum allocation limits.
Institutional investment managers are mainly concerned with liabilities that are 10 to 15 years in the future. But the same investment managers have also to be concerned about the fact that their performance is calculated on a daily basis and that they are likely to be called to account at least every quarter if not every month. Even if a manager is still on track to achieve a 10-year target, they might not survive that long if in the short term they suffer losses that their pension fund client is unwilling to bear. This path-dependent problem can be solved by introducing technology from the world of option pricing, and it takes the form of likelihood-of-loss estimates. By incorporating such estimates into the optimal portfolio allocation exercise, an investment manager can control their expected short-term loss threshold.
Execution used to be conducted through intermediaries whose role was to provide liquidity and immediacy and to aid price discovery. But intermediaries have been dispensed with where internet technology has been deployed to create buy-side-to-buy-side markets. These are especially effective where end investors have at least as much information about their assets as the intermediaries. Advances in execution technology have also allowed smaller institutional investors to leverage their portfolio allocation expertise without having to maintain expensive execution and back-office infrastructure. For example, an investment manager working alone for a self-managed pension fund is now able to execute a fairly sophisticated global bond and equity portfolio allocation complete with an optimised currency hedge, by investing directly in a selection of global index funds and in a selection of currency forwards. The technology that smoothes the path to fast execution also delivers almost instant post-trade confirmation of positions, replacing drifts of illegible trade confirmation faxes with trade and position reports accessed on a secure website.
The final stage of the investment cycle is the measurement of return actually earned and risk actually incurred. This provides the investment manager with the feedback that enables them both to measure the accuracy of their forecasts and to measure their current risk and return position against their targets. Good risk measurement technology must be able to record every single exposure and position, value them and report both absolute and marginal values at every possible level of aggregation and grouping. Absolute and marginal value-at-risk numbers inform the portfolio manager how much influence their current holdings have on their total risk. The same data warehousing technology that allows custodian institutions to provide high-quality risk measurement and reporting also enables these institutions to gain a unique global view of the ebb and flow of institutional investor assets around the world’s capital markets – knowledge that contributes to risk management.
Developments in MPT will continue to keep it at the forefront of financial risk management technology for institutional investors. But what of new departures and of investment science that has not yet become technology? Single period optimisation has the obvious limitation that it does not admit to the existence of intra-period risk, nor look beyond the horizon of the current period. This is a curious situation for a pension fund manager who has to manage to a horizon that is 10 to 15 years in the future, and yet has to use monthly or quarterly optimisation periods. The science and the technology to study optimal allocation problems in continuous time have existed for as long as the famous Black-Scholes solution and are now appearing alongside MPT. The reference earlier to the method of estimating likelihood-of-loss is an example of this confluence of technologies. It is only a matter of time before continuous time optimal portfolio allocation will become part of the standard tool kit of the institutional investor.
It frees every specialist to focus on providing the best service possible to specialists in other areas. It frees investment managers to develop investment strategies and tactics without being concerned about in-house execution, settlement and performance measurement, which will be taken on by back-office outsourcing specialists who can offer economies of scale. Specialism will flourish within investment management. More often than ever, your competitor in one area is your client in another and your service provider in yet another. We are not there yet, but the incentive to get there is strong and clear and exists for all.
Sri Moorthy is a director of State Street Associates in London